智能弹药编队在协同化作战体系中具有潜在应用前景。对于领从架构编队，领弹失效后编队控制缺乏鲁棒性是限制其发展的瓶颈之一。本项目聚焦于研究领弹失效后的智能弹药编队协同控制问题，首先针对智能弹药编队协同控制过程具有马尔科夫性质的特点，通过Q-learning算法短时在线学习获取编队协同控制序贯决策；其次利用小脑模型关节控制器神经网络泛化加速编队执行动作值函数迭代，并根据经验共享原则在编队内部适时共享最优“状态—动作”对，以提高算法收敛速度；最后在此基础上，考虑领弹失效的情况，从弹导引信息获取来源由领弹转换为从弹低成本制导系统，序贯决策先验知识结合从弹短时在线学习构成从弹执行动作优选机制，实现智能弹药编队在状态环境改变后的协同控制。本项目研究旨在提升智能弹药编队的战场环境自适应能力和控制鲁棒性，探明编队最优执行动作序贯决策短时在线生成机理，为基于领弹失效考量的领从架构编队协同控制研究提供理论支撑。

Intelligent ammunition formation has potential application prospect in collaborative combat system. The lack of robustness for formation control is one of the drawbacks restricting the development of formation control after the failure of leader. This project focuses on the collaborative control of intelligent ammunition formation after the failure of leader. Firstly, in view of the markov nature of the collaborative control process of intelligent ammunition formation, the sequential decision of cooperative control of intelligent ammunition formation is obtained by Q-learning algorithm through short-term online learning. Secondly, the neural network of the cerebellar model articulation controller is used to accelerate the iteration of the action value function of the formation, and the optimal "state-action" pairs are timely shared in the formation according to the experience sharing principle, so as to improve the convergence speed of the algorithm. Finally, considering the failure of formation leader, the source of followers guidance information is converted from leader to followers low-cost guidance system. And the sequential decision-making prior knowledge combined with the followers short-term online learning constitutes the optimal choosing mechanism of followers’ action, in order to realize the cooperative control of intelligent ammunition formation after the state environment is changed. The purpose of this project is to improve the adaptive ability and control robustness of intelligent ammunition formation in battlefield environment, and to explore the short-term online generation mechanism of sequential decision-making of optimal execution of formation, so as to provide theoretical support for the research on cooperative control of formation based on the failure consideration of leader.